feat: market regime, self-learning journal, AI investment committee

- regime.py: computed SPY/VIX/sector-momentum snapshot injected into every kickoff
- journal.py: each agent sees its own past picks marked-to-market (learns from wins/losses)
- committee.py: Chief-Risk-Officer LLM pass over the whole book — approve/reduce/veto
  per pick with reasons; memo lands in reports; executor enforces verdicts at open

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016DvdRZkyCy9NJpHYDdPqmS

reporting/journal.py

@@ -0,0 +1,64 @@
+"""Self-learning trade journal — each agent sees its own past picks AND how they worked out
+before researching tonight. Losses and wins become explicit lessons instead of being forgotten.
+
+Joins past theses with their execution price (orders table) and marks each to market via the
+current price. Compact text block injected into the agent's kickoff.
+"""
+
+from __future__ import annotations
+
+from loguru import logger
+
+from config.settings import get_settings
+from state import db as statedb
+from tools.market_data import latest_price
+
+_MAX_ROWS = 10
+
+
+def lessons_for(persona_id: str) -> str:
+    """Past-pick scorecard for one agent. Empty string if no history/DB yet."""
+    s = get_settings()
+    if not s.db_path.exists():
+        return ""
+    try:
+        con = statedb.connect(s.db_path)
+        rows = con.execute(
+            """
+            SELECT t.trade_date, t.ticker, t.direction, t.conviction, t.thesis,
+                   o.est_price, o.status
+            FROM theses t
+            LEFT JOIN orders o
+              ON o.trade_date = t.trade_date AND o.agent = t.agent AND o.ticker = t.ticker
+            WHERE t.agent = ?
+            ORDER BY t.trade_date DESC, t.conviction DESC
+            LIMIT ?
+            """,
+            (persona_id, _MAX_ROWS),
+        ).fetchall()
+        con.close()
+    except Exception as e:
+        logger.warning("journal query failed for {}: {}", persona_id, e)
+        return ""
+    if not rows:
+        return ""
+
+    lines = []
+    for r in rows:
+        entry = r["est_price"]
+        outcome = "not executed"
+        if entry:
+            px = latest_price(r["ticker"])
+            if px:
+                chg = (px / entry - 1) * 100
+                if (r["direction"] or "long") == "short":
+                    chg = -chg
+                outcome = f"{chg:+.1f}% since {r['trade_date']} (entry ~{entry:.2f})"
+        lines.append(
+            f"- {r['trade_date']} {r['ticker']} {r['direction']} conv={r['conviction']}: {outcome}"
+            f" | thesis was: {(r['thesis'] or '')[:100]}"
+        )
+    return (
+        "YOUR TRACK RECORD (mark-to-market of your past picks — learn from it; double down on "
+        "what worked, avoid repeating what failed):\n" + "\n".join(lines)
+    )